Projet 7 : Implémentez un modèle de scoring

Notebook du choix du modèle de scoring

Import

import os

# check scikit-learn version
import sklearn
print(sklearn.__version__)

1.1.1

%pylab inline
import pandas as pd
import seaborn as sns

Populating the interactive namespace from numpy and matplotlib

Configuration

%load_ext autoreload
%autoreload 2

# configurations pandas
pd.set_option('display.max_rows', 250)
pd.set_option('display.max_columns', 250)

import warnings
warnings.filterwarnings("ignore")

Chargement des données

path = './data'
filename = 'Home_Credit_Default_Risk.csv'
data = pd.read_csv(os.path.join(path, filename))
print(data.shape)

(356255, 425)

data_sampling = data.sample(frac=0.1, random_state=42)

Test de LightGBM (Kernel Kagle)

import gc
from sklearn.model_selection import KFold, StratifiedKFold
from lightgbm import LGBMClassifier, early_stopping, log_evaluation
from sklearn.metrics import roc_auc_score

path = './data'
test_filename = 'Test_Home_Credit_Default_Risk.csv'
image_filename = 'Feature_Importance_LightGBM.png'

# Display/plot feature importance
def display_importances(feature_importance_df_, debug=False):
    cols = feature_importance_df_[["feature", "importance"]].groupby("feature").mean().sort_values(by="importance", ascending=False)[:40].index
    best_features = feature_importance_df_.loc[feature_importance_df_.feature.isin(cols)]
    plt.figure(figsize=(8, 10))
    sns.barplot(x="importance", y="feature", data=best_features.sort_values(by="importance", ascending=False), order=cols)
    plt.title('LightGBM Features (avg over folds)')
    plt.tight_layout()
    if not debug:
        plt.savefig(os.path.join(path, image_filename))

# LightGBM GBDT with KFold or Stratified KFold
# Parameters from Tilii kernel: https://www.kaggle.com/tilii7/olivier-lightgbm-parameters-by-bayesian-opt/code
def kfold_lightgbm(df, num_folds, stratified=False, debug=False):
    # Divide in training/validation and test data
    train_df = df[df['TARGET'].notnull()].copy()
    test_df = df[df['TARGET'].isnull()].copy()
    print("Starting LightGBM. Train shape: {}, test shape: {}".format(train_df.shape, test_df.shape))
    del df
    gc.collect()

    # Cross validation model
    if stratified:
        folds = StratifiedKFold(n_splits=num_folds, shuffle=True, random_state=1001)
    else:
        folds = KFold(n_splits=num_folds, shuffle=True, random_state=1001)

    # Create arrays and dataframes to store results
    oof_preds = np.zeros(train_df.shape[0])
    sub_preds = np.zeros(test_df.shape[0])
    feature_importance_df = pd.DataFrame()
    feats = [f for f in train_df.columns if f not in ['TARGET', 'SK_ID_CURR', 'SK_ID_BUREAU', 'SK_ID_PREV', 'index']]

    for n_fold, (train_idx, valid_idx) in enumerate(folds.split(train_df[feats], train_df['TARGET'])):
        train_x, train_y = train_df[feats].iloc[train_idx], train_df['TARGET'].iloc[train_idx]
        valid_x, valid_y = train_df[feats].iloc[valid_idx], train_df['TARGET'].iloc[valid_idx]

        # LightGBM parameters found by Bayesian optimization
        clf = LGBMClassifier(
            n_jobs=8,
            n_estimators=10000,
            learning_rate=0.02,
            num_leaves=34,
            colsample_bytree=0.6,  # 0.9497036,
            subsample=0.8715623,
            max_depth=8,
            reg_alpha=0.041545473,
            reg_lambda=0.0735294,
            min_split_gain=0.0222415,
            min_child_weight=39.3259775,
            verbose=-1,
            force_row_wise=True
        )

        log_evaluation_clb = log_evaluation(period=200)
        early_stopping_clb = early_stopping(stopping_rounds=200)
        clf.fit(
            train_x,
            train_y,
            eval_set=[(train_x, train_y), (valid_x, valid_y)],
            eval_metric='auc',
            callbacks=[early_stopping_clb, log_evaluation_clb]
        )

        oof_preds[valid_idx] = clf.predict_proba(valid_x, num_iteration=clf.best_iteration_)[:, 1]
        sub_preds += clf.predict_proba(test_df[feats], num_iteration=clf.best_iteration_)[:, 1] / folds.n_splits

        fold_importance_df = pd.DataFrame()
        fold_importance_df["feature"] = feats
        fold_importance_df["importance"] = clf.feature_importances_
        fold_importance_df["fold"] = n_fold + 1
        feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)
        print('Fold %2d AUC : %.6f' % (n_fold + 1, roc_auc_score(valid_y, oof_preds[valid_idx])))
        del clf, train_x, train_y, valid_x, valid_y
        gc.collect()

    print('Full AUC score %.6f' % roc_auc_score(train_df['TARGET'], oof_preds))

    # Write submission file and plot feature importance
    if not debug:
        test_df['TARGET'] = sub_preds
        test_df[['SK_ID_CURR', 'TARGET']].to_csv(os.path.join(path, test_filename), index=False)

    display_importances(feature_importance_df, debug=debug)
    return feature_importance_df

feature_importance_df = kfold_lightgbm(data_sampling, num_folds=4, stratified=False, debug=True)

Starting LightGBM. Train shape: (30846, 425), test shape: (4780, 425)
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.867334	training's binary_logloss: 0.216124	valid_1's auc: 0.743922	valid_1's binary_logloss: 0.252913
[400]	training's auc: 0.910685	training's binary_logloss: 0.193713	valid_1's auc: 0.745769	valid_1's binary_logloss: 0.251898
Early stopping, best iteration is:
[312]	training's auc: 0.8937	training's binary_logloss: 0.202581	valid_1's auc: 0.746136	valid_1's binary_logloss: 0.251818
Fold  1 AUC : 0.746136
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.86504	training's binary_logloss: 0.216146	valid_1's auc: 0.744107	valid_1's binary_logloss: 0.253102
[400]	training's auc: 0.907207	training's binary_logloss: 0.194778	valid_1's auc: 0.746499	valid_1's binary_logloss: 0.25165
[600]	training's auc: 0.935452	training's binary_logloss: 0.178235	valid_1's auc: 0.74586	valid_1's binary_logloss: 0.251971
Early stopping, best iteration is:
[421]	training's auc: 0.910349	training's binary_logloss: 0.19296	valid_1's auc: 0.746778	valid_1's binary_logloss: 0.251573
Fold  2 AUC : 0.746778
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.863944	training's binary_logloss: 0.218271	valid_1's auc: 0.736032	valid_1's binary_logloss: 0.246888
[400]	training's auc: 0.905849	training's binary_logloss: 0.196022	valid_1's auc: 0.740528	valid_1's binary_logloss: 0.245304
Early stopping, best iteration is:
[340]	training's auc: 0.895625	training's binary_logloss: 0.201644	valid_1's auc: 0.741379	valid_1's binary_logloss: 0.245162
Fold  3 AUC : 0.741379
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.868906	training's binary_logloss: 0.21716	valid_1's auc: 0.744629	valid_1's binary_logloss: 0.246811
[400]	training's auc: 0.909643	training's binary_logloss: 0.194964	valid_1's auc: 0.746159	valid_1's binary_logloss: 0.245923
Early stopping, best iteration is:
[303]	training's auc: 0.892462	training's binary_logloss: 0.204365	valid_1's auc: 0.746827	valid_1's binary_logloss: 0.245702
Fold  4 AUC : 0.746827
Full AUC score 0.745167

feature_importance_df = kfold_lightgbm(data_sampling, num_folds=4, stratified=True, debug=True)

Starting LightGBM. Train shape: (30846, 425), test shape: (4780, 425)
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.862919	training's binary_logloss: 0.218029	valid_1's auc: 0.749127	valid_1's binary_logloss: 0.248615
[400]	training's auc: 0.907815	training's binary_logloss: 0.195258	valid_1's auc: 0.754375	valid_1's binary_logloss: 0.246677
[600]	training's auc: 0.933878	training's binary_logloss: 0.179564	valid_1's auc: 0.75204	valid_1's binary_logloss: 0.247327
Early stopping, best iteration is:
[423]	training's auc: 0.911753	training's binary_logloss: 0.193093	valid_1's auc: 0.754479	valid_1's binary_logloss: 0.246666
Fold  1 AUC : 0.754479
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.868786	training's binary_logloss: 0.215662	valid_1's auc: 0.728946	valid_1's binary_logloss: 0.253133
[400]	training's auc: 0.908833	training's binary_logloss: 0.193882	valid_1's auc: 0.730834	valid_1's binary_logloss: 0.252502
Early stopping, best iteration is:
[336]	training's auc: 0.897974	training's binary_logloss: 0.199893	valid_1's auc: 0.731192	valid_1's binary_logloss: 0.25245
Fold  2 AUC : 0.731192
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.865982	training's binary_logloss: 0.216867	valid_1's auc: 0.732128	valid_1's binary_logloss: 0.251552
[400]	training's auc: 0.908711	training's binary_logloss: 0.194518	valid_1's auc: 0.734308	valid_1's binary_logloss: 0.250999
Early stopping, best iteration is:
[343]	training's auc: 0.898829	training's binary_logloss: 0.199948	valid_1's auc: 0.73482	valid_1's binary_logloss: 0.250765
Fold  3 AUC : 0.734820
Training until validation scores don't improve for 200 rounds
[200]	training's auc: 0.867075	training's binary_logloss: 0.216943	valid_1's auc: 0.747648	valid_1's binary_logloss: 0.248731
[400]	training's auc: 0.909817	training's binary_logloss: 0.194981	valid_1's auc: 0.750619	valid_1's binary_logloss: 0.247275
Early stopping, best iteration is:
[388]	training's auc: 0.907569	training's binary_logloss: 0.196162	valid_1's auc: 0.750923	valid_1's binary_logloss: 0.247181
Fold  4 AUC : 0.750923
Full AUC score 0.742910

Fonction communes

from lightgbm import LGBMClassifier
from sklearn.model_selection import GridSearchCV
from collections import Counter

def train_test_split(_df):
    # Divide in training/validation and test data
    _train_df = _df[_df['TARGET'].notnull()]
    _test_df = _df[_df['TARGET'].isnull()]
    print("Train shape: {}, test shape: {}".format(_train_df.shape, _test_df.shape))
    counter = Counter(_train_df['TARGET'])
    print("Conteur de class", counter)
    return _train_df, _test_df

def feats_target_split(_df):
    feats = [f for f in _df.columns if f not in ['TARGET', 'SK_ID_CURR', 'SK_ID_BUREAU', 'SK_ID_PREV', 'index']]
    _X = _df[feats].copy()
    _y = _df['TARGET'].copy()
    return _X, _y

def gscv_classifier(_X, _y, _gridParams, _clf):
    print("Starting {}".format(_clf.__class__.__name__))
    _grid = GridSearchCV(_clf, _gridParams, scoring='roc_auc', verbose=2, cv=4)

    # Run the grid
    _grid.fit(_X, _y)

    # Print the best parameters found
    print("Best params:", _grid.best_params_)
    print("Best score:", _grid.best_score_)

    return _grid

from sklearn.metrics import auc
from sklearn.metrics import RocCurveDisplay, roc_auc_score
from sklearn.model_selection import KFold, StratifiedKFold
from xgboost import XGBClassifier
from lightgbm import LGBMClassifier

def _plot_classifier(X, y, clf, title=None, folds=5, stratified=True):
    print("Ploting {}".format(clf.__class__.__name__))
    
    if stratified:
        cv = StratifiedKFold(n_splits=folds, shuffle=True, random_state=1)
    else:
        cv = KFold(n_splits=folds, shuffle=True, random_state=1)

    tprs = []
    aucs = []
    mean_fpr = np.linspace(0, 1, 100)
    oof_preds = np.zeros(X.shape[0])
    feature_importance_df = pd.DataFrame()

    fig, axs = plt.subplots(2, figsize=(15, 20))
    for i, (train, test) in enumerate(cv.split(X, y)):
        train_x, train_y = X.iloc[train], y.iloc[train]
        valid_x, valid_y = X.iloc[test], y.iloc[test]

        clf.fit(train_x, train_y)
        viz = RocCurveDisplay.from_estimator(clf, valid_x, valid_y, name="ROC fold {}".format(i), alpha=0.3, lw=1, ax=axs[1])
        interp_tpr = np.interp(mean_fpr, viz.fpr, viz.tpr)
        interp_tpr[0] = 0.0
        tprs.append(interp_tpr)
        aucs.append(viz.roc_auc)

        if isinstance(clf, XGBClassifier):
            oof_preds[test] = clf.predict_proba(valid_x, iteration_range=(0, clf.best_iteration))[:, 1]
        elif isinstance(clf, LGBMClassifier):
            oof_preds[test] = clf.predict_proba(valid_x, num_iteration=clf.best_iteration_)[:, 1]
        else:
            oof_preds[test] = clf.predict_proba(valid_x)[:, 1]
        fold_importance_df = pd.DataFrame()
        fold_importance_df["feature"] = X.columns
        fold_importance_df["importance"] = clf.feature_importances_
        fold_importance_df["fold"] = i + 1
        feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)
        print('Fold %2d AUC : %.6f' % (i + 1, roc_auc_score(valid_y, oof_preds[test])))

    cols = feature_importance_df[["feature", "importance"]].groupby("feature").mean().sort_values(by="importance", ascending=False)[:20].index
    best_features = feature_importance_df.loc[feature_importance_df.feature.isin(cols)]
    sns.barplot(x="importance", y="feature", data=best_features.sort_values(by="importance", ascending=False), order=cols, ax=axs[0])
    axs[0].set_title('{} Feature Importance (avg over folds)'.format(clf.__class__.__name__))

    axs[1].plot([0, 1], [0, 1], linestyle="--", lw=2, color="r", label="Chance", alpha=0.8)

    mean_tpr = np.mean(tprs, axis=0)
    mean_tpr[-1] = 1.0
    mean_auc = auc(mean_fpr, mean_tpr)
    std_auc = np.std(aucs)
    axs[1].plot(mean_fpr, mean_tpr, color="b", label=r"Mean ROC (AUC = %0.2f $\pm$ %0.2f)" % (mean_auc, std_auc), lw=2, alpha=0.8)

    std_tpr = np.std(tprs, axis=0)
    tprs_upper = np.minimum(mean_tpr + std_tpr, 1)
    tprs_lower = np.maximum(mean_tpr - std_tpr, 0)
    axs[1].fill_between(mean_fpr, tprs_lower, tprs_upper, color="grey", alpha=0.2, label=r"$\pm$ 1 std. dev.")

    axs[1].set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05], title="{} ROC Curve".format(clf.__class__.__name__))
    axs[1].legend(loc="lower right")

    if title:
        plt.suptitle(title, y=1, fontweight ="bold")
    fig.tight_layout()
    plt.show()

from sklearn.metrics import auc
from sklearn.metrics import RocCurveDisplay, roc_auc_score
from sklearn.model_selection import KFold, StratifiedKFold
from xgboost import XGBClassifier
from lightgbm import LGBMClassifier
from imblearn.over_sampling import SMOTE
from imblearn.under_sampling import RandomUnderSampler
from imblearn.pipeline import Pipeline

def plot_balanced_classifier(_X, _y, clf, title=None, folds=5, stratified=True, oversample=False, overparams={}, undersample=False, underparams={}):
    print("Ploting {}".format(clf.__class__.__name__))
    
    if stratified:
        cv = StratifiedKFold(n_splits=folds, shuffle=True, random_state=1)
    else:
        cv = KFold(n_splits=folds, shuffle=True, random_state=1)

    tprs = []
    aucs = []
    mean_fpr = np.linspace(0, 1, 100)
    oof_preds = np.zeros(_X.shape[0])
    feature_importance_df = pd.DataFrame()

    fig, axs = plt.subplots(2, figsize=(15, 20))
    for i, (train_idx, test_idx) in enumerate(cv.split(X, y)):
        _train_x, _train_y = _X.iloc[train_idx], _y.iloc[train_idx]
        _valid_x, _valid_y = _X.iloc[test_idx], _y.iloc[test_idx]

        if oversample and not undersample:
            print("Oversampling...")
            over = SMOTE(**overparams)
            _train_xx, _train_yy = over.fit_resample(_train_x, _train_y)
        elif undersample and not oversample:
            print("Undersampling...")
            under = RandomUnderSampler(**underparams)
            _train_xx, _train_yy = under.fit_resample(_train_x, _train_y)
        elif oversample and undersample:
            print("Pipeline Oversampling and Undersampling...")
            over = SMOTE(**overparams)
            under = RandomUnderSampler(**underparams)
            steps = [('o', over), ('u', under)]
            pipeline = Pipeline(steps=steps)
            _train_xx, _train_yy = pipeline.fit_resample(_train_x, _train_y)
        else:
            _train_xx, _train_yy = _train_x, _train_y

        counter = Counter(_train_yy)
        print("Conteur de class, fold {}: {}".format(i, counter))

        clf.fit(_train_xx, _train_yy)
        viz = RocCurveDisplay.from_estimator(clf, _valid_x, _valid_y, name="ROC fold {}".format(i), alpha=0.3, lw=1, ax=axs[1])
        interp_tpr = np.interp(mean_fpr, viz.fpr, viz.tpr)
        interp_tpr[0] = 0.0
        tprs.append(interp_tpr)
        aucs.append(viz.roc_auc)

        if isinstance(clf, XGBClassifier):
            oof_preds[test_idx] = clf.predict_proba(_valid_x, iteration_range=(0, clf.best_iteration))[:, 1]
        elif isinstance(clf, LGBMClassifier):
            oof_preds[test_idx] = clf.predict_proba(_valid_x, num_iteration=clf.best_iteration_)[:, 1]
        else:
            oof_preds[test_idx] = clf.predict_proba(_valid_x)[:, 1]
        fold_importance_df = pd.DataFrame()
        fold_importance_df["feature"] = X.columns
        fold_importance_df["importance"] = clf.feature_importances_
        fold_importance_df["fold"] = i + 1
        feature_importance_df = pd.concat([feature_importance_df, fold_importance_df], axis=0)
        print('Fold %2d AUC : %.6f' % (i + 1, roc_auc_score(_valid_y, oof_preds[test_idx])))

    cols = feature_importance_df[["feature", "importance"]].groupby("feature").mean().sort_values(by="importance", ascending=False)[:20].index
    best_features = feature_importance_df.loc[feature_importance_df.feature.isin(cols)]
    sns.barplot(x="importance", y="feature", data=best_features.sort_values(by="importance", ascending=False), order=cols, ax=axs[0])
    axs[0].set_title('{} Feature Importance (avg over folds)'.format(clf.__class__.__name__))

    axs[1].plot([0, 1], [0, 1], linestyle="--", lw=2, color="r", label="Chance", alpha=0.8)

    mean_tpr = np.mean(tprs, axis=0)
    mean_tpr[-1] = 1.0
    mean_auc = auc(mean_fpr, mean_tpr)
    std_auc = np.std(aucs)
    axs[1].plot(mean_fpr, mean_tpr, color="b", label=r"Mean ROC (AUC = %0.2f $\pm$ %0.2f)" % (mean_auc, std_auc), lw=2, alpha=0.8)

    std_tpr = np.std(tprs, axis=0)
    tprs_upper = np.minimum(mean_tpr + std_tpr, 1)
    tprs_lower = np.maximum(mean_tpr - std_tpr, 0)
    axs[1].fill_between(mean_fpr, tprs_lower, tprs_upper, color="grey", alpha=0.2, label=r"$\pm$ 1 std. dev.")

    axs[1].set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05], title="{} ROC Curve".format(clf.__class__.__name__))
    axs[1].legend(loc="lower right")

    if title:
        plt.suptitle(title, y=1, fontweight ="bold")
    fig.tight_layout()
    plt.show()

    return clf

Test de LightGBM (GridSearchCV)

from lightgbm import LGBMClassifier

gridParams = {
    'num_leaves': [8, 16, 34],  # Shall be smaller than 2^(max_depth)
    'max_depth': [8],
    'min_child_samples': [100, 1000],
    'learning_rate': [0.01, 0.05, 0.1],
    'colsample_bytree': [0.1, 0.3, 0.5],  # Alias for feature_fraction
}

# Initiate classifier to use
clf = LGBMClassifier( 
    objective='binary', 
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
lgbm_grid = gscv_classifier(X, y, gridParams, clf)
_ = plot_balanced_classifier(X, y, lgbm_grid.best_estimator_, "LightGBM without balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting LGBMClassifier
Fitting 4 folds for each of 54 candidates, totalling 216 fits
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.9s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   2.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.9s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.9s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.8s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
Best params: {'colsample_bytree': 0.3, 'learning_rate': 0.05, 'max_depth': 8, 'min_child_samples': 1000, 'num_leaves': 34}
Best score: 0.7399814473499138
Ploting LGBMClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.757400
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.747618
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.731920
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.738462
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.740896

from lightgbm import LGBMClassifier

gridParams = {
    'num_leaves': [8, 16, 34],  # Shall be smaller than 2^(max_depth)
    'max_depth': [8],
    'min_child_samples': [100, 1000],
    'learning_rate': [0.01, 0.05, 0.1],
    'colsample_bytree': [0.1, 0.3, 0.5],  # Alias for feature_fraction
}

# Initiate classifier to use
clf = LGBMClassifier( 
    objective='binary', 
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
lgbm_grid = gscv_classifier(X, y, gridParams, clf)
plot_balanced_classifier(X, y, lgbm_grid.best_estimator_, "LightGBM with balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting LGBMClassifier
Fitting 4 folds for each of 54 candidates, totalling 216 fits
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   0.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   0.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.5s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   0.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.8s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.9s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.8s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.01, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.8s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.05, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.0s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.5s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=16; total time=   1.4s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=100, num_leaves=34; total time=   1.6s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=8; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.3s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=16; total time=   1.1s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.7s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.2s
[CV] END colsample_bytree=0.5, learning_rate=0.1, max_depth=8, min_child_samples=1000, num_leaves=34; total time=   1.3s
Best params: {'colsample_bytree': 0.5, 'learning_rate': 0.1, 'max_depth': 8, 'min_child_samples': 1000, 'num_leaves': 16}
Best score: 0.7390624475743153
Ploting LGBMClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.754718
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.743641
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.727650
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.750862
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.730055

LGBMClassifier(colsample_bytree=0.5, force_row_wise=True, is_unbalance=True,
               max_depth=8, min_child_samples=1000, n_jobs=8, num_leaves=16,
               objective='binary', verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Test de XGBoost (GridSearchCV)

from xgboost import XGBClassifier

gridParams = {
    'learning_rate': [0.01, 0.1],
    'max_depth': [4, 6],
    'subsample': [0.6, 0.9],
    'colsample_bytree': [0.1, 0.3],
    'n_estimators': [400, 800]
}

# Initiate classifier to use
clf = XGBClassifier( 
    verbosity=0,
    objective='binary:logistic',
    nthread=8,
    use_label_encoder=False,
    error_score='raise',
)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
xgb_grid = gscv_classifier(X, y, gridParams, clf)
plot_balanced_classifier(X, y, xgb_grid.best_estimator_, "XGBoost without balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting XGBClassifier
Fitting 4 folds for each of 32 candidates, totalling 128 fits
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  10.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.3s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.3s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.3s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  23.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  23.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  23.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   8.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.3s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.2s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.2s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.3s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.3s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.5s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.2s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.5s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  23.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.5s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  20.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  16.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  31.6s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  33.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  33.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  34.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  32.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  22.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  23.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  23.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  22.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  22.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  22.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  22.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  43.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  42.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  47.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  45.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  44.6s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  44.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  44.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  43.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  16.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  16.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  15.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  29.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  29.6s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  29.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  30.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  29.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  29.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  29.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  29.3s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  20.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  22.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  20.6s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  41.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  43.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  41.9s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  46.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  44.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  44.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  46.6s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  45.1s
Best params: {'colsample_bytree': 0.3, 'learning_rate': 0.01, 'max_depth': 6, 'n_estimators': 800, 'subsample': 0.6}
Best score: 0.743837713501315
Ploting XGBClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.759556
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.743982
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.732931
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.743384
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.747262

XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,
              colsample_bynode=1, colsample_bytree=0.3,
              enable_categorical=False, error_score='raise', gamma=0, gpu_id=-1,
              importance_type=None, interaction_constraints='',
              learning_rate=0.01, max_delta_step=0, max_depth=6,
              min_child_weight=1, missing=nan, monotone_constraints='()',
              n_estimators=800, n_jobs=8, nthread=8, num_parallel_tree=1,
              predictor='auto', random_state=0, reg_alpha=0, reg_lambda=1,
              scale_pos_weight=1, subsample=0.6, tree_method='exact',
              use_label_encoder=False, validate_parameters=1, ...)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

from xgboost import XGBClassifier

gridParams = {
    'learning_rate': [0.01, 0.1],
    'max_depth': [4, 6],
    'subsample': [0.6, 0.9],
    'colsample_bytree': [0.1, 0.3],
    'n_estimators': [400, 800]
}

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
positive_class = y.sum()
negatvie_class = len(y) - positive_class
scale_pos_weight = negatvie_class / positive_class

# Initiate classifier to use
clf = XGBClassifier( 
    verbosity=0,
    objective='binary:logistic',
    nthread=8,
    use_label_encoder=False,
    error_score='raise',
    scale_pos_weight=scale_pos_weight,
)

xgb_grid = gscv_classifier(X, y, gridParams, clf)
plot_balanced_classifier(X, y, xgb_grid.best_estimator_, "XGBoost with balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting XGBClassifier
Fitting 4 folds for each of 32 candidates, totalling 128 fits
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  10.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=   9.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.0s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.3s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  18.3s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.7s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  12.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.5s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  12.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  12.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  12.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.8s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.2s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.6s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.9s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  23.1s
[CV] END colsample_bytree=0.1, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  23.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=   8.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   9.2s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=   8.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  17.2s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  16.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.3s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.0s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  17.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  16.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.4s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  11.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.6s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  11.7s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.5s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.8s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  21.9s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.1s
[CV] END colsample_bytree=0.1, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  22.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.6s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.6; total time=  16.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  16.8s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  32.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  31.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  33.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.6; total time=  33.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  33.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  31.3s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  31.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=4, n_estimators=800, subsample=0.9; total time=  32.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  22.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  21.6s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  21.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.6; total time=  21.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.5s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=400, subsample=0.9; total time=  21.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  42.7s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  42.6s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  41.4s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.6; total time=  40.2s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  39.9s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  41.0s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  40.1s
[CV] END colsample_bytree=0.3, learning_rate=0.01, max_depth=6, n_estimators=800, subsample=0.9; total time=  39.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.6; total time=  14.4s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  15.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.1s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=400, subsample=0.9; total time=  14.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  27.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.6; total time=  28.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  27.6s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  27.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  27.3s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=4, n_estimators=800, subsample=0.9; total time=  30.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  20.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  19.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.6; total time=  19.3s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  19.5s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  19.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  20.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=400, subsample=0.9; total time=  19.3s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  38.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  38.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  40.8s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.6; total time=  39.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  38.0s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  37.7s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  38.2s
[CV] END colsample_bytree=0.3, learning_rate=0.1, max_depth=6, n_estimators=800, subsample=0.9; total time=  37.9s
Best params: {'colsample_bytree': 0.3, 'learning_rate': 0.01, 'max_depth': 4, 'n_estimators': 800, 'subsample': 0.6}
Best score: 0.7401563916288427
Ploting XGBClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.752596
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.743611
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.732443
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.737996
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.740938

XGBClassifier(base_score=0.5, booster='gbtree', colsample_bylevel=1,
              colsample_bynode=1, colsample_bytree=0.3,
              enable_categorical=False, error_score='raise', gamma=0, gpu_id=-1,
              importance_type=None, interaction_constraints='',
              learning_rate=0.01, max_delta_step=0, max_depth=4,
              min_child_weight=1, missing=nan, monotone_constraints='()',
              n_estimators=800, n_jobs=8, nthread=8, num_parallel_tree=1,
              predictor='auto', random_state=0, reg_alpha=0, reg_lambda=1,
              scale_pos_weight=11.503445480340494, subsample=0.6,
              tree_method='exact', use_label_encoder=False,
              validate_parameters=1, ...)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Test de RandomForest (GridSearchCV)

from sklearn.ensemble import RandomForestClassifier

gridParams = {
    'n_estimators': [100, 400, 700],
    'max_depth': [3, 4, 5, 6],
    'max_samples': [0.5, 1],
}

# Initiate classifier to use
clf = RandomForestClassifier(
    n_jobs=8,
    verbose=0,
)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
rfc_grid = gscv_classifier(X, y, gridParams, clf)
plot_balanced_classifier(X, y, rfc_grid.best_estimator_, "RandomForest without balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting RandomForestClassifier
Fitting 4 folds for each of 24 candidates, totalling 96 fits
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=  23.4s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   0.8s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   0.8s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   0.7s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.2s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.1s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.1s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.2s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.4s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.4s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.4s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.7s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.2s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   0.9s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   0.9s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   0.9s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.8s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.5s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.6s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.6s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.3s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.2s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.1s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.2s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.2s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.5s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.0s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   4.8s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.0s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   4.9s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.5s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.4s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.8s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.5s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   5.5s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   5.6s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   5.7s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   5.7s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.7s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.2s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.2s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.3s
Best params: {'max_depth': 6, 'max_samples': 0.5, 'n_estimators': 700}
Best score: 0.7092335672546197
Ploting RandomForestClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.723066
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.716841
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.696013
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.707200
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.713347

RandomForestClassifier(max_depth=6, max_samples=0.5, n_estimators=700, n_jobs=8)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

from sklearn.ensemble import RandomForestClassifier

gridParams = {
    'n_estimators': [100, 400, 700],
    'max_depth': [3, 4, 5, 6],
    'max_samples': [0.5, 1],
}

# Initiate classifier to use
clf = RandomForestClassifier(
    n_jobs=8,
    class_weight='balanced',
    verbose=0,
)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
rfc_grid = gscv_classifier(X, y, gridParams, clf)
plot_balanced_classifier(X, y, rfc_grid.best_estimator_, "RandomForest with balanced class weight")

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Starting RandomForestClassifier
Fitting 4 folds for each of 24 candidates, totalling 96 fits
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   0.9s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.3s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.2s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.3s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=400; total time=   2.3s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.9s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.5s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.5s
[CV] END .....max_depth=3, max_samples=0.5, n_estimators=700; total time=   3.5s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=3, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   1.1s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   1.0s
[CV] END .......max_depth=3, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=3, max_samples=1, n_estimators=700; total time=   1.5s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=100; total time=   1.0s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   3.4s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.7s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.7s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=400; total time=   2.9s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.3s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.4s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.6s
[CV] END .....max_depth=4, max_samples=0.5, n_estimators=700; total time=   4.5s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=4, max_samples=1, n_estimators=400; total time=   1.1s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.6s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.5s
[CV] END .......max_depth=4, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=100; total time=   1.1s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.2s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.3s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=400; total time=   3.4s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.3s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.4s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.6s
[CV] END .....max_depth=5, max_samples=0.5, n_estimators=700; total time=   5.3s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=5, max_samples=1, n_estimators=400; total time=   1.1s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   2.1s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.5s
[CV] END .......max_depth=5, max_samples=1, n_estimators=700; total time=   1.6s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.2s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=100; total time=   1.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.9s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.6s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.6s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=400; total time=   3.7s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   6.2s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   6.3s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   6.2s
[CV] END .....max_depth=6, max_samples=0.5, n_estimators=700; total time=   6.0s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.5s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=100; total time=   0.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.9s
[CV] END .......max_depth=6, max_samples=1, n_estimators=400; total time=   0.8s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.3s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.6s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.4s
[CV] END .......max_depth=6, max_samples=1, n_estimators=700; total time=   1.3s
Best params: {'max_depth': 5, 'max_samples': 0.5, 'n_estimators': 400}
Best score: 0.7154566051827315
Ploting RandomForestClassifier
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 1973})
Fold  1 AUC : 0.728331
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 1974})
Fold  2 AUC : 0.715927
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 1974})
Fold  3 AUC : 0.703136
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 1974})
Fold  4 AUC : 0.709835
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 1973})
Fold  5 AUC : 0.721471

RandomForestClassifier(class_weight='balanced', max_depth=5, max_samples=0.5,
                       n_estimators=400, n_jobs=8)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Over-sampling

import imblearn
print(imblearn.__version__)

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)

0.7.0
Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})

from collections import Counter
from imblearn.over_sampling import SMOTE

# summarize class distribution
counter = Counter(y)
print("Before resampling:", counter)
print("Ratio of positive class: {}".format(counter[1] / len(y)))

# transform the dataset
over = SMOTE()
# transform the dataset
oX, oy = over.fit_resample(X, y)
# summarize the new class distribution
counter = Counter(oy)
print(counter)

Before resampling: Counter({0.0: 28379, 1.0: 2467})
Ratio of positive class: 0.07997795500226934
Counter({0.0: 28379, 1.0: 28379})

from lightgbm import LGBMClassifier

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
plot_balanced_classifier(X, y, clf, "LightGBM with oversampling", oversample=True)

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Ploting LGBMClassifier
Oversampling...
Conteur de class, fold 0: Counter({0.0: 22703, 1.0: 22703})
Fold  1 AUC : 0.719533
Oversampling...
Conteur de class, fold 1: Counter({0.0: 22703, 1.0: 22703})
Fold  2 AUC : 0.695094
Oversampling...
Conteur de class, fold 2: Counter({0.0: 22703, 1.0: 22703})
Fold  3 AUC : 0.705016
Oversampling...
Conteur de class, fold 3: Counter({0.0: 22703, 1.0: 22703})
Fold  4 AUC : 0.718867
Oversampling...
Conteur de class, fold 4: Counter({0.0: 22704, 1.0: 22704})
Fold  5 AUC : 0.711065

LGBMClassifier(colsample_bytree=0.5, force_row_wise=True, is_unbalance=True,
               max_depth=8, min_child_samples=1000, n_jobs=8, num_leaves=16,
               objective='binary', verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Under-sampling

import imblearn
print(imblearn.__version__)

train_df, _ = train_test_split(data)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)

0.7.0
Train shape: (307511, 425), test shape: (48744, 425)
Conteur de class Counter({0.0: 282686, 1.0: 24825})

from collections import Counter
from imblearn.under_sampling import RandomUnderSampler

# summarize class distribution
counter = Counter(y)
print("Before resampling:", counter)
print("Ratio of positive class: {}".format(counter[1] / len(y)))

# transform the dataset
under = RandomUnderSampler()
# transform the dataset
X, y = under.fit_resample(X, y)
# summarize the new class distribution
counter = Counter(y)
print(counter)

Before resampling: Counter({0.0: 282686, 1.0: 24825})
Ratio of positive class: 0.08072881945686496
Counter({0.0: 24825, 1.0: 24825})

from lightgbm import LGBMClassifier

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
plot_balanced_classifier(X, y, clf, "LightGBM with undersampling", undersample=True)

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Ploting LGBMClassifier
Undersampling...
Conteur de class, fold 0: Counter({0.0: 1973, 1.0: 1973})
Fold  1 AUC : 0.734981
Undersampling...
Conteur de class, fold 1: Counter({0.0: 1974, 1.0: 1974})
Fold  2 AUC : 0.717625
Undersampling...
Conteur de class, fold 2: Counter({0.0: 1974, 1.0: 1974})
Fold  3 AUC : 0.707004
Undersampling...
Conteur de class, fold 3: Counter({0.0: 1974, 1.0: 1974})
Fold  4 AUC : 0.725478
Undersampling...
Conteur de class, fold 4: Counter({0.0: 1973, 1.0: 1973})
Fold  5 AUC : 0.719662

LGBMClassifier(colsample_bytree=0.5, force_row_wise=True, is_unbalance=True,
               max_depth=8, min_child_samples=1000, n_jobs=8, num_leaves=16,
               objective='binary', verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Mixed Over/Under-sampling

import imblearn
print(imblearn.__version__)

train_df, _ = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)

0.7.0
Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})

from collections import Counter
from imblearn.over_sampling import SMOTE
from imblearn.under_sampling import RandomUnderSampler
from imblearn.pipeline import Pipeline

# summarize class distribution
counter = Counter(y)
print("Before resampling:", counter)
print("Ratio of positive class: {}".format(counter[1] / len(y)))

# transform the dataset
over = SMOTE(sampling_strategy=0.1)
under = RandomUnderSampler(sampling_strategy=0.5)
steps = [('o', over), ('u', under)]
pipeline = Pipeline(steps=steps)
# transform the dataset
X, y = pipeline.fit_resample(X, y)
# summarize the new class distribution
counter = Counter(y)
print(counter)

Before resampling: Counter({0.0: 28379, 1.0: 2467})
Ratio of positive class: 0.07997795500226934
Counter({0.0: 5674, 1.0: 2837})

from lightgbm import LGBMClassifier

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X.fillna(X.median(), inplace=True)
plot_balanced_classifier(
    X,
    y,
    clf,
    "LightGBM with under/over-sampling",
    undersample=True,
    oversample=True,
    underparams={'sampling_strategy':0.5},
    overparams={'sampling_strategy':0.1}
)

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})
Ploting LGBMClassifier
Pipeline Oversampling and Undersampling...
Conteur de class, fold 0: Counter({0.0: 4540, 1.0: 2270})
Fold  1 AUC : 0.744904
Pipeline Oversampling and Undersampling...
Conteur de class, fold 1: Counter({0.0: 4540, 1.0: 2270})
Fold  2 AUC : 0.728864
Pipeline Oversampling and Undersampling...
Conteur de class, fold 2: Counter({0.0: 4540, 1.0: 2270})
Fold  3 AUC : 0.722399
Pipeline Oversampling and Undersampling...
Conteur de class, fold 3: Counter({0.0: 4540, 1.0: 2270})
Fold  4 AUC : 0.736994
Pipeline Oversampling and Undersampling...
Conteur de class, fold 4: Counter({0.0: 4540, 1.0: 2270})
Fold  5 AUC : 0.728496

LGBMClassifier(colsample_bytree=0.5, force_row_wise=True, is_unbalance=True,
               max_depth=8, min_child_samples=1000, n_jobs=8, num_leaves=16,
               objective='binary', verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

comment sortir un modèle entrainé à partir des folds ?

Explication avec LIME

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X_test, y_test = feats_target_split(test_df)
X.fillna(X.median(), inplace=True)
X_test.fillna(X_test.median(), inplace=True)

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})

from lime import lime_tabular
from lightgbm import LGBMClassifier

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

clf.fit(X, y, eval_set=[(X, y)], eval_metric='auc', early_stopping_rounds=10)

explainer = lime_tabular.LimeTabularExplainer(
    X.values,
    mode="classification",
    class_names=["0", "1"],
    feature_names=X.columns.to_list(),
)

idx = random.randint(1, len(X_test))

explanation = explainer.explain_instance(X_test.iloc[idx], clf.predict_proba, num_features=len(X.columns))

explanation.show_in_notebook()

[1]	training's auc: 0.656564	training's binary_logloss: 0.284852
[2]	training's auc: 0.688334	training's binary_logloss: 0.302254
[3]	training's auc: 0.728079	training's binary_logloss: 0.318197
[4]	training's auc: 0.742795	training's binary_logloss: 0.335407
[5]	training's auc: 0.750071	training's binary_logloss: 0.352901
[6]	training's auc: 0.755264	training's binary_logloss: 0.369976
[7]	training's auc: 0.758383	training's binary_logloss: 0.386172
[8]	training's auc: 0.762716	training's binary_logloss: 0.402352
[9]	training's auc: 0.765007	training's binary_logloss: 0.416585
[10]	training's auc: 0.767662	training's binary_logloss: 0.429761
[11]	training's auc: 0.768917	training's binary_logloss: 0.442517
[12]	training's auc: 0.771207	training's binary_logloss: 0.45399
[13]	training's auc: 0.773099	training's binary_logloss: 0.464611
[14]	training's auc: 0.77497	training's binary_logloss: 0.473975
[15]	training's auc: 0.777909	training's binary_logloss: 0.482927
[16]	training's auc: 0.780359	training's binary_logloss: 0.49094
[17]	training's auc: 0.783231	training's binary_logloss: 0.498095
[18]	training's auc: 0.785193	training's binary_logloss: 0.504342
[19]	training's auc: 0.787953	training's binary_logloss: 0.509577
[20]	training's auc: 0.789888	training's binary_logloss: 0.514736
[21]	training's auc: 0.790688	training's binary_logloss: 0.51927
[22]	training's auc: 0.792694	training's binary_logloss: 0.522947
[23]	training's auc: 0.79387	training's binary_logloss: 0.526251
[24]	training's auc: 0.794977	training's binary_logloss: 0.529357
[25]	training's auc: 0.796639	training's binary_logloss: 0.532183
[26]	training's auc: 0.79839	training's binary_logloss: 0.534443
[27]	training's auc: 0.800353	training's binary_logloss: 0.536274
[28]	training's auc: 0.80164	training's binary_logloss: 0.537989
[29]	training's auc: 0.803062	training's binary_logloss: 0.539332
[30]	training's auc: 0.804005	training's binary_logloss: 0.540502
[31]	training's auc: 0.804856	training's binary_logloss: 0.541432
[32]	training's auc: 0.805929	training's binary_logloss: 0.54211
[33]	training's auc: 0.807423	training's binary_logloss: 0.542511
[34]	training's auc: 0.808132	training's binary_logloss: 0.542973
[35]	training's auc: 0.809185	training's binary_logloss: 0.543211
[36]	training's auc: 0.810041	training's binary_logloss: 0.54337
[37]	training's auc: 0.811658	training's binary_logloss: 0.543183
[38]	training's auc: 0.812664	training's binary_logloss: 0.543266
[39]	training's auc: 0.813886	training's binary_logloss: 0.543039
[40]	training's auc: 0.81498	training's binary_logloss: 0.542707
[41]	training's auc: 0.816247	training's binary_logloss: 0.542364
[42]	training's auc: 0.817277	training's binary_logloss: 0.542048
[43]	training's auc: 0.817942	training's binary_logloss: 0.541588
[44]	training's auc: 0.818958	training's binary_logloss: 0.541065
[45]	training's auc: 0.820182	training's binary_logloss: 0.540341
[46]	training's auc: 0.821349	training's binary_logloss: 0.539638
[47]	training's auc: 0.822321	training's binary_logloss: 0.539236
[48]	training's auc: 0.823113	training's binary_logloss: 0.538717
[49]	training's auc: 0.823987	training's binary_logloss: 0.537926
[50]	training's auc: 0.825181	training's binary_logloss: 0.5372
[51]	training's auc: 0.825858	training's binary_logloss: 0.536425
[52]	training's auc: 0.82695	training's binary_logloss: 0.535682
[53]	training's auc: 0.827954	training's binary_logloss: 0.534648
[54]	training's auc: 0.828947	training's binary_logloss: 0.533798
[55]	training's auc: 0.830061	training's binary_logloss: 0.532843
[56]	training's auc: 0.830547	training's binary_logloss: 0.532163
[57]	training's auc: 0.831424	training's binary_logloss: 0.531308
[58]	training's auc: 0.832662	training's binary_logloss: 0.530164
[59]	training's auc: 0.833343	training's binary_logloss: 0.529481
[60]	training's auc: 0.834185	training's binary_logloss: 0.528634
[61]	training's auc: 0.835042	training's binary_logloss: 0.527733
[62]	training's auc: 0.836457	training's binary_logloss: 0.526536
[63]	training's auc: 0.837284	training's binary_logloss: 0.525765
[64]	training's auc: 0.838172	training's binary_logloss: 0.52486
[65]	training's auc: 0.839109	training's binary_logloss: 0.52405
[66]	training's auc: 0.840048	training's binary_logloss: 0.523147
[67]	training's auc: 0.840672	training's binary_logloss: 0.522322
[68]	training's auc: 0.841547	training's binary_logloss: 0.521412
[69]	training's auc: 0.842452	training's binary_logloss: 0.520488
[70]	training's auc: 0.843243	training's binary_logloss: 0.519572
[71]	training's auc: 0.843589	training's binary_logloss: 0.518952
[72]	training's auc: 0.844457	training's binary_logloss: 0.518109
[73]	training's auc: 0.845404	training's binary_logloss: 0.517161
[74]	training's auc: 0.846149	training's binary_logloss: 0.516298
[75]	training's auc: 0.847158	training's binary_logloss: 0.515343
[76]	training's auc: 0.847909	training's binary_logloss: 0.514555
[77]	training's auc: 0.848792	training's binary_logloss: 0.513656
[78]	training's auc: 0.849698	training's binary_logloss: 0.512762
[79]	training's auc: 0.850395	training's binary_logloss: 0.511906
[80]	training's auc: 0.851204	training's binary_logloss: 0.51096
[81]	training's auc: 0.851926	training's binary_logloss: 0.510069
[82]	training's auc: 0.852584	training's binary_logloss: 0.509186
[83]	training's auc: 0.853022	training's binary_logloss: 0.508719
[84]	training's auc: 0.85364	training's binary_logloss: 0.507918
[85]	training's auc: 0.854563	training's binary_logloss: 0.507031
[86]	training's auc: 0.855288	training's binary_logloss: 0.506195
[87]	training's auc: 0.856089	training's binary_logloss: 0.505316
[88]	training's auc: 0.856777	training's binary_logloss: 0.50449
[89]	training's auc: 0.857656	training's binary_logloss: 0.503528
[90]	training's auc: 0.858267	training's binary_logloss: 0.502989
[91]	training's auc: 0.859027	training's binary_logloss: 0.502142
[92]	training's auc: 0.859978	training's binary_logloss: 0.50127
[93]	training's auc: 0.860731	training's binary_logloss: 0.5004
[94]	training's auc: 0.861708	training's binary_logloss: 0.499515
[95]	training's auc: 0.862075	training's binary_logloss: 0.499073
[96]	training's auc: 0.862918	training's binary_logloss: 0.498172
[97]	training's auc: 0.863562	training's binary_logloss: 0.497374
[98]	training's auc: 0.864385	training's binary_logloss: 0.496486
[99]	training's auc: 0.86492	training's binary_logloss: 0.495818
[100]	training's auc: 0.8658	training's binary_logloss: 0.494867

Explication avec SHAP

import shap
# print the JS visualization code to the notebook
shap.initjs()

train_df, test_df = train_test_split(data_sampling)
X, y = feats_target_split(train_df)
X_test, y_test = feats_target_split(test_df)
X.fillna(X.median(), inplace=True)
X_test.fillna(X_test.median(), inplace=True)

Train shape: (30846, 425), test shape: (4780, 425)
Conteur de class Counter({0.0: 28379, 1.0: 2467})

from lightgbm import LGBMClassifier
import shap

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

clf.fit(X, y, eval_set=[(X, y)], eval_metric='auc', early_stopping_rounds=10)

explainer = shap.TreeExplainer(clf)
shap_values = explainer.shap_values(X)
sv = np.array(shap_values)
y = clf.predict(X).astype("bool")
sv_positive = sv[:,y,:]
sv_negative = sv[:,~y,:]

[1]	training's auc: 0.656564	training's binary_logloss: 0.284852
[2]	training's auc: 0.688334	training's binary_logloss: 0.302254
[3]	training's auc: 0.728079	training's binary_logloss: 0.318197
[4]	training's auc: 0.742795	training's binary_logloss: 0.335407
[5]	training's auc: 0.750071	training's binary_logloss: 0.352901
[6]	training's auc: 0.755264	training's binary_logloss: 0.369976
[7]	training's auc: 0.758383	training's binary_logloss: 0.386172
[8]	training's auc: 0.762716	training's binary_logloss: 0.402352
[9]	training's auc: 0.765007	training's binary_logloss: 0.416585
[10]	training's auc: 0.767662	training's binary_logloss: 0.429761
[11]	training's auc: 0.768917	training's binary_logloss: 0.442517
[12]	training's auc: 0.771207	training's binary_logloss: 0.45399
[13]	training's auc: 0.773099	training's binary_logloss: 0.464611
[14]	training's auc: 0.77497	training's binary_logloss: 0.473975
[15]	training's auc: 0.777909	training's binary_logloss: 0.482927
[16]	training's auc: 0.780359	training's binary_logloss: 0.49094
[17]	training's auc: 0.783231	training's binary_logloss: 0.498095
[18]	training's auc: 0.785193	training's binary_logloss: 0.504342
[19]	training's auc: 0.787953	training's binary_logloss: 0.509577
[20]	training's auc: 0.789888	training's binary_logloss: 0.514736
[21]	training's auc: 0.790688	training's binary_logloss: 0.51927
[22]	training's auc: 0.792694	training's binary_logloss: 0.522947
[23]	training's auc: 0.79387	training's binary_logloss: 0.526251
[24]	training's auc: 0.794977	training's binary_logloss: 0.529357
[25]	training's auc: 0.796639	training's binary_logloss: 0.532183
[26]	training's auc: 0.79839	training's binary_logloss: 0.534443
[27]	training's auc: 0.800353	training's binary_logloss: 0.536274
[28]	training's auc: 0.80164	training's binary_logloss: 0.537989
[29]	training's auc: 0.803062	training's binary_logloss: 0.539332
[30]	training's auc: 0.804005	training's binary_logloss: 0.540502
[31]	training's auc: 0.804856	training's binary_logloss: 0.541432
[32]	training's auc: 0.805929	training's binary_logloss: 0.54211
[33]	training's auc: 0.807423	training's binary_logloss: 0.542511
[34]	training's auc: 0.808132	training's binary_logloss: 0.542973
[35]	training's auc: 0.809185	training's binary_logloss: 0.543211
[36]	training's auc: 0.810041	training's binary_logloss: 0.54337
[37]	training's auc: 0.811658	training's binary_logloss: 0.543183
[38]	training's auc: 0.812664	training's binary_logloss: 0.543266
[39]	training's auc: 0.813886	training's binary_logloss: 0.543039
[40]	training's auc: 0.81498	training's binary_logloss: 0.542707
[41]	training's auc: 0.816247	training's binary_logloss: 0.542364
[42]	training's auc: 0.817277	training's binary_logloss: 0.542048
[43]	training's auc: 0.817942	training's binary_logloss: 0.541588
[44]	training's auc: 0.818958	training's binary_logloss: 0.541065
[45]	training's auc: 0.820182	training's binary_logloss: 0.540341
[46]	training's auc: 0.821349	training's binary_logloss: 0.539638
[47]	training's auc: 0.822321	training's binary_logloss: 0.539236
[48]	training's auc: 0.823113	training's binary_logloss: 0.538717
[49]	training's auc: 0.823987	training's binary_logloss: 0.537926
[50]	training's auc: 0.825181	training's binary_logloss: 0.5372
[51]	training's auc: 0.825858	training's binary_logloss: 0.536425
[52]	training's auc: 0.82695	training's binary_logloss: 0.535682
[53]	training's auc: 0.827954	training's binary_logloss: 0.534648
[54]	training's auc: 0.828947	training's binary_logloss: 0.533798
[55]	training's auc: 0.830061	training's binary_logloss: 0.532843
[56]	training's auc: 0.830547	training's binary_logloss: 0.532163
[57]	training's auc: 0.831424	training's binary_logloss: 0.531308
[58]	training's auc: 0.832662	training's binary_logloss: 0.530164
[59]	training's auc: 0.833343	training's binary_logloss: 0.529481
[60]	training's auc: 0.834185	training's binary_logloss: 0.528634
[61]	training's auc: 0.835042	training's binary_logloss: 0.527733
[62]	training's auc: 0.836457	training's binary_logloss: 0.526536
[63]	training's auc: 0.837284	training's binary_logloss: 0.525765
[64]	training's auc: 0.838172	training's binary_logloss: 0.52486
[65]	training's auc: 0.839109	training's binary_logloss: 0.52405
[66]	training's auc: 0.840048	training's binary_logloss: 0.523147
[67]	training's auc: 0.840672	training's binary_logloss: 0.522322
[68]	training's auc: 0.841547	training's binary_logloss: 0.521412
[69]	training's auc: 0.842452	training's binary_logloss: 0.520488
[70]	training's auc: 0.843243	training's binary_logloss: 0.519572
[71]	training's auc: 0.843589	training's binary_logloss: 0.518952
[72]	training's auc: 0.844457	training's binary_logloss: 0.518109
[73]	training's auc: 0.845404	training's binary_logloss: 0.517161
[74]	training's auc: 0.846149	training's binary_logloss: 0.516298
[75]	training's auc: 0.847158	training's binary_logloss: 0.515343
[76]	training's auc: 0.847909	training's binary_logloss: 0.514555
[77]	training's auc: 0.848792	training's binary_logloss: 0.513656
[78]	training's auc: 0.849698	training's binary_logloss: 0.512762
[79]	training's auc: 0.850395	training's binary_logloss: 0.511906
[80]	training's auc: 0.851204	training's binary_logloss: 0.51096
[81]	training's auc: 0.851926	training's binary_logloss: 0.510069
[82]	training's auc: 0.852584	training's binary_logloss: 0.509186
[83]	training's auc: 0.853022	training's binary_logloss: 0.508719
[84]	training's auc: 0.85364	training's binary_logloss: 0.507918
[85]	training's auc: 0.854563	training's binary_logloss: 0.507031
[86]	training's auc: 0.855288	training's binary_logloss: 0.506195
[87]	training's auc: 0.856089	training's binary_logloss: 0.505316
[88]	training's auc: 0.856777	training's binary_logloss: 0.50449
[89]	training's auc: 0.857656	training's binary_logloss: 0.503528
[90]	training's auc: 0.858267	training's binary_logloss: 0.502989
[91]	training's auc: 0.859027	training's binary_logloss: 0.502142
[92]	training's auc: 0.859978	training's binary_logloss: 0.50127
[93]	training's auc: 0.860731	training's binary_logloss: 0.5004
[94]	training's auc: 0.861708	training's binary_logloss: 0.499515
[95]	training's auc: 0.862075	training's binary_logloss: 0.499073
[96]	training's auc: 0.862918	training's binary_logloss: 0.498172
[97]	training's auc: 0.863562	training's binary_logloss: 0.497374
[98]	training's auc: 0.864385	training's binary_logloss: 0.496486
[99]	training's auc: 0.86492	training's binary_logloss: 0.495818
[100]	training's auc: 0.8658	training's binary_logloss: 0.494867

shap.summary_plot(shap_values[1], X.astype("float"))

shap.dependence_plot("EXT_SOURCE_2", shap_values[1], X)

shap.dependence_plot("CODE_GENDER", shap_values[1], X)

shap.dependence_plot("DAYS_BIRTH", shap_values[1], X)

Optimisation du modèle

from sklearn.model_selection import train_test_split as sk_train_test_split
train_df, test_df = train_test_split(data)
X, y = feats_target_split(train_df)
X_train, X_test, y_train, y_test = sk_train_test_split(X, y)

Train shape: (307511, 425), test shape: (48744, 425)
Conteur de class Counter({0.0: 282686, 1.0: 24825})

# Minimiser les faux négatifs
from lightgbm import LGBMClassifier

# Initiate classifier to use
clf = LGBMClassifier(
    objective='binary',
    n_jobs=8,
    verbose=-1,
    force_row_wise=True,
    is_unbalance=True,  # try to Automatically balance the weight of the dominated labels
    num_leaves=16,  # Shall be smaller than 2^(max_depth)
    max_depth=8,
    min_child_samples=1000,
    learning_rate=0.1,
    colsample_bytree=0.5,  # Alias for feature_fraction
)

# X_train.fillna(X_train.median(), inplace=True)
clf.fit(X_train, y_train, eval_set=[(X_train, y_train)], eval_metric='auc', early_stopping_rounds=10)

[1]	training's auc: 0.63919	training's binary_logloss: 0.28723
[2]	training's auc: 0.696242	training's binary_logloss: 0.302327
[3]	training's auc: 0.708291	training's binary_logloss: 0.32089
[4]	training's auc: 0.717796	training's binary_logloss: 0.341245
[5]	training's auc: 0.721938	training's binary_logloss: 0.360294
[6]	training's auc: 0.723682	training's binary_logloss: 0.378874
[7]	training's auc: 0.72552	training's binary_logloss: 0.396688
[8]	training's auc: 0.728493	training's binary_logloss: 0.414097
[9]	training's auc: 0.72961	training's binary_logloss: 0.429746
[10]	training's auc: 0.731878	training's binary_logloss: 0.44479
[11]	training's auc: 0.734915	training's binary_logloss: 0.458422
[12]	training's auc: 0.736239	training's binary_logloss: 0.471378
[13]	training's auc: 0.737125	training's binary_logloss: 0.482759
[14]	training's auc: 0.737768	training's binary_logloss: 0.493301
[15]	training's auc: 0.739295	training's binary_logloss: 0.503274
[16]	training's auc: 0.739661	training's binary_logloss: 0.512071
[17]	training's auc: 0.741247	training's binary_logloss: 0.520292
[18]	training's auc: 0.741786	training's binary_logloss: 0.52756
[19]	training's auc: 0.742639	training's binary_logloss: 0.534071
[20]	training's auc: 0.742958	training's binary_logloss: 0.540166
[21]	training's auc: 0.743664	training's binary_logloss: 0.545591
[22]	training's auc: 0.744574	training's binary_logloss: 0.55043
[23]	training's auc: 0.745367	training's binary_logloss: 0.554765
[24]	training's auc: 0.746243	training's binary_logloss: 0.558735
[25]	training's auc: 0.746828	training's binary_logloss: 0.562303
[26]	training's auc: 0.747931	training's binary_logloss: 0.565409
[27]	training's auc: 0.748314	training's binary_logloss: 0.568307
[28]	training's auc: 0.749331	training's binary_logloss: 0.570763
[29]	training's auc: 0.750022	training's binary_logloss: 0.572904
[30]	training's auc: 0.750785	training's binary_logloss: 0.574908
[31]	training's auc: 0.751522	training's binary_logloss: 0.576542
[32]	training's auc: 0.75248	training's binary_logloss: 0.577889
[33]	training's auc: 0.753101	training's binary_logloss: 0.579207
[34]	training's auc: 0.753965	training's binary_logloss: 0.580317
[35]	training's auc: 0.754494	training's binary_logloss: 0.581272
[36]	training's auc: 0.755323	training's binary_logloss: 0.581955
[37]	training's auc: 0.755951	training's binary_logloss: 0.582612
[38]	training's auc: 0.756461	training's binary_logloss: 0.583277
[39]	training's auc: 0.757106	training's binary_logloss: 0.583668
[40]	training's auc: 0.758043	training's binary_logloss: 0.583681
[41]	training's auc: 0.758576	training's binary_logloss: 0.584016
[42]	training's auc: 0.759043	training's binary_logloss: 0.584172
[43]	training's auc: 0.759703	training's binary_logloss: 0.584294
[44]	training's auc: 0.760128	training's binary_logloss: 0.584415
[45]	training's auc: 0.760824	training's binary_logloss: 0.584356
[46]	training's auc: 0.761319	training's binary_logloss: 0.584362
[47]	training's auc: 0.761861	training's binary_logloss: 0.584286
[48]	training's auc: 0.762391	training's binary_logloss: 0.584165
[49]	training's auc: 0.762871	training's binary_logloss: 0.584019
[50]	training's auc: 0.763345	training's binary_logloss: 0.583863
[51]	training's auc: 0.763871	training's binary_logloss: 0.583667
[52]	training's auc: 0.764403	training's binary_logloss: 0.583398
[53]	training's auc: 0.764771	training's binary_logloss: 0.583203
[54]	training's auc: 0.765118	training's binary_logloss: 0.582984
[55]	training's auc: 0.765679	training's binary_logloss: 0.58269
[56]	training's auc: 0.766493	training's binary_logloss: 0.581969
[57]	training's auc: 0.766895	training's binary_logloss: 0.58171
[58]	training's auc: 0.767398	training's binary_logloss: 0.581396
[59]	training's auc: 0.767757	training's binary_logloss: 0.581115
[60]	training's auc: 0.768176	training's binary_logloss: 0.580823
[61]	training's auc: 0.768644	training's binary_logloss: 0.580471
[62]	training's auc: 0.769001	training's binary_logloss: 0.580194
[63]	training's auc: 0.769432	training's binary_logloss: 0.579881
[64]	training's auc: 0.769831	training's binary_logloss: 0.579578
[65]	training's auc: 0.770289	training's binary_logloss: 0.579194
[66]	training's auc: 0.770599	training's binary_logloss: 0.578885
[67]	training's auc: 0.771015	training's binary_logloss: 0.578503
[68]	training's auc: 0.771333	training's binary_logloss: 0.578228
[69]	training's auc: 0.77164	training's binary_logloss: 0.577962
[70]	training's auc: 0.771985	training's binary_logloss: 0.577648
[71]	training's auc: 0.772285	training's binary_logloss: 0.577363
[72]	training's auc: 0.772745	training's binary_logloss: 0.576955
[73]	training's auc: 0.773055	training's binary_logloss: 0.576658
[74]	training's auc: 0.773359	training's binary_logloss: 0.576383
[75]	training's auc: 0.773719	training's binary_logloss: 0.576041
[76]	training's auc: 0.774045	training's binary_logloss: 0.575749
[77]	training's auc: 0.774308	training's binary_logloss: 0.57548
[78]	training's auc: 0.774638	training's binary_logloss: 0.57515
[79]	training's auc: 0.775002	training's binary_logloss: 0.574828
[80]	training's auc: 0.775299	training's binary_logloss: 0.574493
[81]	training's auc: 0.775609	training's binary_logloss: 0.574199
[82]	training's auc: 0.775872	training's binary_logloss: 0.57389
[83]	training's auc: 0.776177	training's binary_logloss: 0.573611
[84]	training's auc: 0.776437	training's binary_logloss: 0.573345
[85]	training's auc: 0.776768	training's binary_logloss: 0.57305
[86]	training's auc: 0.777064	training's binary_logloss: 0.57277
[87]	training's auc: 0.777333	training's binary_logloss: 0.572516
[88]	training's auc: 0.777649	training's binary_logloss: 0.572201
[89]	training's auc: 0.778001	training's binary_logloss: 0.571892
[90]	training's auc: 0.778252	training's binary_logloss: 0.571613
[91]	training's auc: 0.778634	training's binary_logloss: 0.571237
[92]	training's auc: 0.778901	training's binary_logloss: 0.570969
[93]	training's auc: 0.779209	training's binary_logloss: 0.570686
[94]	training's auc: 0.779518	training's binary_logloss: 0.570371
[95]	training's auc: 0.779781	training's binary_logloss: 0.570107
[96]	training's auc: 0.779976	training's binary_logloss: 0.569892
[97]	training's auc: 0.780247	training's binary_logloss: 0.569653
[98]	training's auc: 0.780429	training's binary_logloss: 0.56942
[99]	training's auc: 0.780727	training's binary_logloss: 0.569159
[100]	training's auc: 0.780939	training's binary_logloss: 0.568939

LGBMClassifier(colsample_bytree=0.5, force_row_wise=True, is_unbalance=True,
               max_depth=8, min_child_samples=1000, n_jobs=8, num_leaves=16,
               objective='binary', verbose=-1)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

predproba = clf.predict_proba(X_test)
predprobapos = predproba[:,1]
print("Proba:", predprobapos)
auc_score = roc_auc_score(y_test, predprobapos)
print("AUC:", auc_score)

Proba: [0.18511533 0.33047083 0.41513223 ... 0.21674607 0.64368859 0.48100921]
AUC: 0.7574614918450082

# minimiser les faux negatifs en pondérant FN et FP
from sklearn.metrics import confusion_matrix
metric_metier = {}
threshold = np.linspace(0,1,101)
for i in threshold:
    print(">> Threshold", i)
    predclass = predprobapos.copy()
    predclass[predclass < i] = 0
    predclass[predclass >= i] = 1
    res = confusion_matrix(y_test, predclass)
    tn, fp, fn, tp = res.ravel()
    print("True negative:", tn)
    print("False positive:", fp)
    metric_metier[i] = (10*fn+fp)
key_min = min(metric_metier, key=metric_metier.get)
print("------------------------------------------------------")
print("Minimum:", key_min)
print("Minimum value:", metric_metier[key_min])

>> Threshold 0.0
True negative: 0
False positive: 70598
>> Threshold 0.01
True negative: 0
False positive: 70598
>> Threshold 0.02
True negative: 0
False positive: 70598
>> Threshold 0.03
True negative: 3
False positive: 70595
>> Threshold 0.04
True negative: 45
False positive: 70553
>> Threshold 0.05
True negative: 125
False positive: 70473
>> Threshold 0.06
True negative: 284
False positive: 70314
>> Threshold 0.07
True negative: 513
False positive: 70085
>> Threshold 0.08
True negative: 824
False positive: 69774
>> Threshold 0.09
True negative: 1234
False positive: 69364
>> Threshold 0.1
True negative: 1690
False positive: 68908
>> Threshold 0.11
True negative: 2234
False positive: 68364
>> Threshold 0.12
True negative: 2850
False positive: 67748
>> Threshold 0.13
True negative: 3586
False positive: 67012
>> Threshold 0.14
True negative: 4429
False positive: 66169
>> Threshold 0.15
True negative: 5280
False positive: 65318
>> Threshold 0.16
True negative: 6267
False positive: 64331
>> Threshold 0.17
True negative: 7359
False positive: 63239
>> Threshold 0.18
True negative: 8412
False positive: 62186
>> Threshold 0.19
True negative: 9618
False positive: 60980
>> Threshold 0.2
True negative: 10825
False positive: 59773
>> Threshold 0.21
True negative: 12056
False positive: 58542
>> Threshold 0.22
True negative: 13286
False positive: 57312
>> Threshold 0.23
True negative: 14607
False positive: 55991
>> Threshold 0.24
True negative: 15870
False positive: 54728
>> Threshold 0.25
True negative: 17197
False positive: 53401
>> Threshold 0.26
True negative: 18595
False positive: 52003
>> Threshold 0.27
True negative: 20043
False positive: 50555
>> Threshold 0.28
True negative: 21432
False positive: 49166
>> Threshold 0.29
True negative: 22785
False positive: 47813
>> Threshold 0.3
True negative: 24136
False positive: 46462
>> Threshold 0.31
True negative: 25523
False positive: 45075
>> Threshold 0.32
True negative: 26936
False positive: 43662
>> Threshold 0.33
True negative: 28260
False positive: 42338
>> Threshold 0.34
True negative: 29695
False positive: 40903
>> Threshold 0.35000000000000003
True negative: 31044
False positive: 39554
>> Threshold 0.36
True negative: 32421
False positive: 38177
>> Threshold 0.37
True negative: 33717
False positive: 36881
>> Threshold 0.38
True negative: 35044
False positive: 35554
>> Threshold 0.39
True negative: 36379
False positive: 34219
>> Threshold 0.4
True negative: 37659
False positive: 32939
>> Threshold 0.41000000000000003
True negative: 38907
False positive: 31691
>> Threshold 0.42
True negative: 40180
False positive: 30418
>> Threshold 0.43
True negative: 41420
False positive: 29178
>> Threshold 0.44
True negative: 42688
False positive: 27910
>> Threshold 0.45
True negative: 43868
False positive: 26730
>> Threshold 0.46
True negative: 44983
False positive: 25615
>> Threshold 0.47000000000000003
True negative: 46127
False positive: 24471
>> Threshold 0.48
True negative: 47322
False positive: 23276
>> Threshold 0.49
True negative: 48390
False positive: 22208
>> Threshold 0.5
True negative: 49440
False positive: 21158
>> Threshold 0.51
True negative: 50495
False positive: 20103
>> Threshold 0.52
True negative: 51535
False positive: 19063
>> Threshold 0.53
True negative: 52567
False positive: 18031
>> Threshold 0.54
True negative: 53514
False positive: 17084
>> Threshold 0.55
True negative: 54455
False positive: 16143
>> Threshold 0.56
True negative: 55391
False positive: 15207
>> Threshold 0.5700000000000001
True negative: 56288
False positive: 14310
>> Threshold 0.58
True negative: 57210
False positive: 13388
>> Threshold 0.59
True negative: 58076
False positive: 12522
>> Threshold 0.6
True negative: 58859
False positive: 11739
>> Threshold 0.61
True negative: 59685
False positive: 10913
>> Threshold 0.62
True negative: 60460
False positive: 10138
>> Threshold 0.63
True negative: 61221
False positive: 9377
>> Threshold 0.64
True negative: 61927
False positive: 8671
>> Threshold 0.65
True negative: 62617
False positive: 7981
>> Threshold 0.66
True negative: 63270
False positive: 7328
>> Threshold 0.67
True negative: 63881
False positive: 6717
>> Threshold 0.68
True negative: 64521
False positive: 6077
>> Threshold 0.6900000000000001
True negative: 65109
False positive: 5489
>> Threshold 0.7000000000000001
True negative: 65626
False positive: 4972
>> Threshold 0.71
True negative: 66133
False positive: 4465
>> Threshold 0.72
True negative: 66596
False positive: 4002
>> Threshold 0.73
True negative: 67024
False positive: 3574
>> Threshold 0.74
True negative: 67491
False positive: 3107
>> Threshold 0.75
True negative: 67903
False positive: 2695
>> Threshold 0.76
True negative: 68235
False positive: 2363
>> Threshold 0.77
True negative: 68565
False positive: 2033
>> Threshold 0.78
True negative: 68862
False positive: 1736
>> Threshold 0.79
True negative: 69178
False positive: 1420
>> Threshold 0.8
True negative: 69421
False positive: 1177
>> Threshold 0.81
True negative: 69671
False positive: 927
>> Threshold 0.8200000000000001
True negative: 69825
False positive: 773
>> Threshold 0.8300000000000001
True negative: 69987
False positive: 611
>> Threshold 0.84
True negative: 70137
False positive: 461
>> Threshold 0.85
True negative: 70257
False positive: 341
>> Threshold 0.86
True negative: 70358
False positive: 240
>> Threshold 0.87
True negative: 70440
False positive: 158
>> Threshold 0.88
True negative: 70490
False positive: 108
>> Threshold 0.89
True negative: 70532
False positive: 66
>> Threshold 0.9
True negative: 70569
False positive: 29
>> Threshold 0.91
True negative: 70587
False positive: 11
>> Threshold 0.92
True negative: 70597
False positive: 1
>> Threshold 0.93
True negative: 70598
False positive: 0
>> Threshold 0.9400000000000001
True negative: 70598
False positive: 0
>> Threshold 0.9500000000000001
True negative: 70598
False positive: 0
>> Threshold 0.96
True negative: 70598
False positive: 0
>> Threshold 0.97
True negative: 70598
False positive: 0
>> Threshold 0.98
True negative: 70598
False positive: 0
>> Threshold 0.99
True negative: 70598
False positive: 0
>> Threshold 1.0
True negative: 70598
False positive: 0
------------------------------------------------------
Minimum: 0.51
Minimum value: 40853

# minimiser les faux negatifs en utilisant le fbeta score
from sklearn.metrics import fbeta_score
metric_metier = {}
threshold = np.linspace(0,1,101)
for i in threshold:
    print(">> Threshold", i)
    predclass = predprobapos.copy()
    predclass[predclass < i] = 0
    predclass[predclass >= i] = 1
    res = fbeta_score(y_test, predclass, beta=2)
    metric_metier[i] = res
key_max = max(metric_metier, key=metric_metier.get)
print("------------------------------------------------------")
print("Maximum:", key_max)
print("Maximum value:", metric_metier[key_max])

>> Threshold 0.0
>> Threshold 0.01
>> Threshold 0.02
>> Threshold 0.03
>> Threshold 0.04
>> Threshold 0.05
>> Threshold 0.06
>> Threshold 0.07
>> Threshold 0.08
>> Threshold 0.09
>> Threshold 0.1
>> Threshold 0.11
>> Threshold 0.12
>> Threshold 0.13
>> Threshold 0.14
>> Threshold 0.15
>> Threshold 0.16
>> Threshold 0.17
>> Threshold 0.18
>> Threshold 0.19
>> Threshold 0.2
>> Threshold 0.21
>> Threshold 0.22
>> Threshold 0.23
>> Threshold 0.24
>> Threshold 0.25
>> Threshold 0.26
>> Threshold 0.27
>> Threshold 0.28
>> Threshold 0.29
>> Threshold 0.3
>> Threshold 0.31
>> Threshold 0.32
>> Threshold 0.33
>> Threshold 0.34
>> Threshold 0.35000000000000003
>> Threshold 0.36
>> Threshold 0.37
>> Threshold 0.38
>> Threshold 0.39
>> Threshold 0.4
>> Threshold 0.41000000000000003
>> Threshold 0.42
>> Threshold 0.43
>> Threshold 0.44
>> Threshold 0.45
>> Threshold 0.46
>> Threshold 0.47000000000000003
>> Threshold 0.48
>> Threshold 0.49
>> Threshold 0.5
>> Threshold 0.51
>> Threshold 0.52
>> Threshold 0.53
>> Threshold 0.54
>> Threshold 0.55
>> Threshold 0.56
>> Threshold 0.5700000000000001
>> Threshold 0.58
>> Threshold 0.59
>> Threshold 0.6
>> Threshold 0.61
>> Threshold 0.62
>> Threshold 0.63
>> Threshold 0.64
>> Threshold 0.65
>> Threshold 0.66
>> Threshold 0.67
>> Threshold 0.68
>> Threshold 0.6900000000000001
>> Threshold 0.7000000000000001
>> Threshold 0.71
>> Threshold 0.72
>> Threshold 0.73
>> Threshold 0.74
>> Threshold 0.75
>> Threshold 0.76
>> Threshold 0.77
>> Threshold 0.78
>> Threshold 0.79
>> Threshold 0.8
>> Threshold 0.81
>> Threshold 0.8200000000000001
>> Threshold 0.8300000000000001
>> Threshold 0.84
>> Threshold 0.85
>> Threshold 0.86
>> Threshold 0.87
>> Threshold 0.88
>> Threshold 0.89
>> Threshold 0.9
>> Threshold 0.91
>> Threshold 0.92
>> Threshold 0.93
>> Threshold 0.9400000000000001
>> Threshold 0.9500000000000001
>> Threshold 0.96
>> Threshold 0.97
>> Threshold 0.98
>> Threshold 0.99
>> Threshold 1.0
------------------------------------------------------
Maximum: 0.49
Maximum value: 0.4255668747946106

Export du notebook en HTML

!jupyter nbconvert --config nbconvert/config_html.py

[NbConvertApp] Converting notebook P7_01_notebookexploration.ipynb to html
[NbConvertApp] Writing 4623390 bytes to P7_01_notebookexploration.html
[NbConvertApp] Converting notebook P7_02_notebookscoring.ipynb to html
[NbConvertApp] Writing 4437532 bytes to P7_02_notebookscoring.html
[NbConvertApp] Converting notebook P7_03_notebookdashboard.ipynb to html
C:\Users\flori\miniconda3\envs\jupyter\lib\site-packages\nbconvert\filters\datatypefilter.py:39: UserWarning: Your element with mimetype(s) dict_keys(['application/vnd.plotly.v1+json']) is not able to be represented.
  warn("Your element with mimetype(s) {mimetypes}"
[NbConvertApp] Writing 901023 bytes to P7_03_notebookdashboard.html